Critical Amino Acids in the Active Site of Meprin Metalloproteinases for Substrate and Peptide Bond Specificity*

The protease domains of the evolutionarily related alpha and beta subunits of meprin metalloproteases are approximately 55% identical at the amino acid level; however, their substrate and peptide bond specificities differ markedly. The meprin beta subunit favors acidic residues proximal to the scissile bond, while the alpha subunit prefers small or aromatic amino acids flanking the scissile bond. Thus gastrin, a peptide that contains a string of five Glu residues, is an excellent substrate for meprin beta, while it is not hydrolyzed by meprin alpha. Work herein aimed to identify critical amino acids in the meprin active sites that determine the substrate specificity differences. Sequence alignments and homology models, based on the crystal structure of the crayfish astacin, showed electrostatic differences within the meprin active sites. Site-directed mutagenesis of active site residues demonstrated that replacement of a hydrophobic residue by a basic amino acid enabled the meprin alpha protease to cleave gastrin. The meprin alphaY199K mutant was most effective; the corresponding mutation of meprin betaK185Y resulted in decreased activity toward gastrin. Peptide cleavage site determinations and kinetic analyses using a variety of peptides extended evidence that meprin alphaTyr-199/betaLys-185 are substrate specificity determinants in meprin active sites. These studies shed light on the molecular basis for the substrate specificity differences of astacin metalloproteinases.


INTRODUCTION
The meprin α and ß subunits are zinc metalloendopeptidases of the 'astacin family' and 'metzincin superfamily' that can form homo-and heterooligomeric complexes (1,2). The subunits derived from a common multidomain ancestor protein, but have evolved to have markedly different substrate and peptide bond specificities, structural properties, chromosomal locations, and membrane associations (3)(4)(5). Meprin ß has a preference for hydrolysis of peptide bonds containing acidic amino acid residues, whereas meprin α prefers to cleave bonds flanked by small or hydrophobic residues. The meprin α subunit has a propensity to form large homooligomeric complexes containing 12 to 100 subunits, while meprin ß alone forms only homodimers (5). Both subunits are synthesized as membrane-spanning type 1 proteins in the endoplasmic reticulum, however, while the meprin ß subunit remains membrane-bound through the secretory pathway and at the plasma membrane, the meprin α subunit is proteolytically processed during biosynthesis and thus loses its transmembrane domain and is secreted unless it associates with meprin ß (6). Meprin α is encoded on chromosome 17 in mice and 6 in humans near the histocompatibility complex; meprin ß is encoded on chromosome 18 in mice and humans (7). While both meprin subunits are expressed in embryonic kidney proximal tubule cells and intestinal epithelial cells, the subunits are expressed differentially postnatally, and one or the other subunit appears to be upregulated in cancer cells (8)(9)(10). Both meprin subunits cleave a variety of peptides and proteins and have a preference for peptides larger than 6 amino acids, indicating extended substrate binding sites (3,11). The best peptide substrates identified for meprin α were gastrin-releasing peptide, cholecystokinin, glucagon, substance P and valosin. The first three of the latter peptides are also substrates for meprin ß, the last two are not (3,4). Gastrin is by far the best substrate identified for meprin ß, and it is not cleaved by mouse meprin α.
Cytokines are also substrates for meprins implicating them in inflammatory and immune system processes (12). For example, monocyte chemoattractant protein-1 (MCP-1) is cleaved by meprin α, while osteopontin is a substrate for meprin ß (3). Azocasein and gelatin are often used as substrates to assay both forms of meprin. Other protein substrates for both include extracellular proteins, such as collagen type IV, laminin 1 and 5, fibronectin, and nidogen. Meprins are known to play a critical role in development, and have been implicated in cancer metastasis, inflammatory bowel disease, and in kidney diseases (8,(12)(13)(14)(15)(16). The increasing knowledge accumulating about the regulation and substrate specificities of these enzymes will help to define the role of these proteinases in physiologic and pathological processes.
The aim of the study herein was to shed light on the molecular basis for the differences in the specificity of the subunits. Homology models and sequence alignments of the meprin active sites were examined for potential residues affecting peptide bond specificity of the subunits. Three amino acid residues in positions within the active sites were identified as potential contributors to the differences in specificities of meprins α trypsin inhibitor (type II-S, Sigma) was added at 2-fold excess over trypsin. As a control, promeprins were incubated with trypsin preincubated with soybean trypsin inhibitor; no meprin proteolytic activity was observed under these conditions. Azocaseinase activity was measured as described previously (18). Kinetic analyses for peptide hydrolysis were 9 determined by quantitative HPLC analysis (3). Cleavage sites were determined as previously described using a Perceptive Biosystems Linear Voyager matrix-assisted laser Homology models of meprin protease domains, based on the crystal structure of astacin, were utilized to determine the positions of the potentially critical basic residues in meprin ß implicated by the sequence alignments ( Fig. 2). Three residues, denoted by arrows, which were electrostatically different in the subunits and located within the active site were proposed to be potentially critical to substrate interactions. The three residues are αF161, αY199, and αP228; the corresponding residues are ßR147, ßK185 and ßK214.
The localization of these basic amino acids in the meprin ß subunit implicates these residues in acidic substrate specificity. The αF161 residue (ßR147) is located at a potentially critical position within the central α-helix containing two of the His zinc ligands and the catalytic Glu residue. Residue αY199 (ßK185) protrudes into the active site from the lower floor of the cleft. Residues αP228 and ßK214, also at the floor of the active site cleft, are located just two residues after a tyrosine residue in the SxMHY sequence of the 'Met turn', a structural feature of astacin proteases involved in transition state stabilization (19). These three potentially critical residues were mutated in mouse α to the corresponding basic meprin ß residue to determine whether they contribute to subunit substrate specificity differences.
General Proteolytic Activity of Mutants -The three mutants, αF161R, αY199K and αP228K, were expressed in 293 cells and purified utilizing a His-tag placed at the mature COOH-terminus (17,21). To determine overall stability and activatibility, the mutant proteins were subjected to proteolysis by trypsin. Each mutant was enzymatically activated and decreased in mass by trypsin to the same extent as wild-type meprin (data not shown), indicating that the prosequence was removed; there was no evidence of extensive degradation of these mutants which has been observed to occur with unstable mutants (22). To determine whether the mutations affected overall proteolytic activity, hydrolysis of the protein substrate azocasein was examined (Table I). Mutants αY199K and αF161R degraded azocasein with specific activity similar to wild-type meprin α, while the αP228K mutant hydrolyzed the protein substrate less efficiently. This decreased activity may be due to an alteration in the orientation of Tyr residue involved in the transition state in the conserved 'Met turn' sequence thus affecting the catalytic efficiency (23,24). Overall, however, these results indicate that the 3 meprin α mutants are proteolytically active and there are no major changes in stability of the proteins to trypsin.

Mutations of Amino Acid Residues Identified Above Enable Meprin α Hydrolysis
of Gastrin 17. Several peptides were incubated with wild-type mouse meprin α, 12 αY199K, αF161R, and αP228K to further examine whether substrate specificity was altered by the mutations (Table II). The substrates tested had previously been determined to be either cleaved by meprin ß only (gastrin and orcokinin), cleaved by both meprin α and ß (cholecystokinin), or cleaved by meprin α only (neurotensin, LHRH, bradykinin).
The most marked affect was that all three meprin α active site mutations were able to cleave the ß-specific peptide gastrin 17. The meprin αY199K mutant was the most effective against gastrin. By contrast, another meprin ß specific peptide, orcokinin, was not cleaved by any of the mutants. Meprin ß has a much greater affinity for gastrin 17 compared to orcokinin (K M of 7 µM compared to 100 µM, respectively), and this might contribute to the differences observed for the meprin α mutants (3). In general, the αY199K mutant was more effective against the other substrates tested (cholecystokinin, neurotensin, LHRH) than the wild-type enzyme or the other mutants. The αP228K mutant either hydrolyzed the peptides to the same extent as the wild-type protein, or it was 2-fold less effective. The αF161R mutant was more effective in cleaving cholecystokinin and less effective in cleaving neurotensin than the wild-type enzyme.
To better determine the interactions of peptide substrates with the mutant meprins, kinetic parameters were determined (Table III) The catalytic effectiveness of αY199K against cholecytokinin was approximately 7-fold greater than the wild-type enzyme and the αP228K mutant and 2-fold greater than αF161R (Table III). All the mutants had lower K M values than the wild-type enzyme, and the αF161R and αY199K mutants showed modest increases in k cat values. The increased catalytic efficiency of the mutants was primarily due to greater affinity for the substrate.
Effects of Mutation of Meprin ßK185Y -Because the αY199K mutant was the most effective in enhancing activity of meprin α towards gastrin and other peptides, the equivalent residue in mepirn ß (ßK185Y) was mutated. The wild-type rat meprin ß and the ßK185Y mutant had similar activity against azocasein (Table IV). The ßK185Y mutant also retained the ability to hydrolyze gastrin and cholecystokinin. In addition, this mutant hydryolyzed αMSH more rapidly than the wild-type meprin ß, and was able to cleave neurotensin unlike the wild-type ß protein.
Mutation of the ßK185 residue to Tyr significantly reduced the specificity constant for gastrin from 105 to 8.9 x 10 5 M -1 s -1 (Table V) The kinetic constants for hydrolysis of CCK8s and αMSH were also determined for meprin ß and the ßK185Y mutant. The amino acid replacement resulted in a 1.6-fold increase in k cat /K M for CCK8s and a 13-fold increase in this value for αMSH. Therefore, while the ßK185Y mutant cleaved gastrin less effectively than the wild-type protein, it was more effective for other substrates.

Cleavage Site Analysis Demonstrates the Changes in Substrate Specificity for the
Active Site Mutants. Determination of cleavage sites within the peptides gastrin, CCK8s and αMSH were determined through the separation of products by HPLC and identification by MALDI-TOF (Table VI). This analysis revealed that the αY199K mutant hydrolyzed some of the same bonds within the gastrin peptide as meprin ß, however, there was a cleavage site COOH-terminal to the Asp residue rather than NH 2 terminal to this residue. The bond COOH-terminal to this Asp residue is also hydrolyzed by the αF161R mutant, but this is the only cleavage site for this mutant.
The cleavage site analysis of CCK8s revealed that meprin α and the three α mutants hydrolyze this peptide between a Trp and Met residue and between an Asp and Phe at the COOH-terminus. Although the cleavage sites remain unchanged, the amount of hydrolysis at each site differs (Fig. 3). There is an approximately equal distribution of CCK8s hydrolysis product peaks for the wild-type meprin α and for the αP228K mutant. amino acid βK185 interacts with substrates at both P1 and P1', while βR147 appears to interact at P1. The αY199K mutant had a much higher Km value for gastrin than did meprin ß, indicating that there are multiple factors in the meprin ß active site that determine affinity. This is consistent with results of previous studies that had shown that meprins have a preference for substrates with a minimum of 6 to 8 amino acids (3,26) indicating an extended binding site.
The substrate specificity of many members of the metzincin superfamily have been studied, and the great majority of these metalloproteinases have preferences for http://www.jbc.org/ Downloaded from this generallity, e.g. aggrecanases cleave at Glu residues (28). Two other members of the astacin family beside meprin ß that cleave peptide bonds flanked by negatively charged amino acids are BMP-1, a mammalian procollagen C-proteinase, and flavastacin, an astacin in flavobacteria (4,29,30). Homology models of the protease domains of these enzymes, implicate equivalent basic amino acids as substrate specificity determinants as in meprin ß (Fig. 3). The models show the presence of prominent basic residues located on the lower active site cleft that resides in the putative primed substrate binding regions.
Meprin α, crayfish astacin and a fish hatching enzyme (HCE-1) prefer small or hydrophobic amino acids at P1' and these peptidases lack the basic residue at the base of the active site cleft (31)(32)(33). Thus, the results of the studies herein with meprins are applicable to other astacins.
Insights into the meprin active site residues that affect affinity for substrates will be applicable to the design of inhibitors of these proteases. Meprins are not inhibited by any known naturally occurring metalloproteinase inhibitors (such as tissue inhibitors of metalloproteinases, TIMPs), and specific inhibitors for the subunits are not yet available (3, 4. 17). Yet these proteinases are induced in several types of cancer cells (e.g., colon and pancreatic cancer) and in leukocytes at inflammatory sites (such as in inflammatory bowel disease) (9,10,12). Chromatograms of CCK8s treated with wild-type merpin α, αY199K, αF161R and αP228K mutants separated on a reverse phase HPLC column. The hydrolysis products were identified by MALDI-TOF and correspond to the W-M and D-F hydrolysis sites.